We've designed the 'Get the Offer' course to further aid your case interview preparations. This includes detailed walkthroughs of 10 more market sizing questions, live interview examples, 40+ case studies, exhibits, and structured drills. Get hands-on experience and ace your interviews! Explore more at: prepmatter.com/get-the-offer.
Great video, helpful content! As someone who's always been interested in management consulting, I found this really informative. Keep up the good work Deniz! I hope I'll get into the BCG Istanbul office this year :)
Yo, this is such an over complicated method. These are simple fermi estimation problems. No interviewer expects such minutiae, they want estimation in the correct order of magnitude. The irony is the mode you detail the the parameters, higher the risk of being wrong.
Great video. I did a different approach for first case. First, I calculated total number of pregnant women at a given time point to be 375K. (6M londoners, number of 21-40yo is 1/4 so around 1.5M, out of which est 25% to be pregnant). Noting that total pregnancy duration as 40 weeks or 280 days, I decided to get rid of that factor entirely because I thought it would be plausible for anticipated mothers to buy their clothing ahead of time to make the total duration 1 year instead of 280 days for ease of math. Multipled by average number of visits to store/year as factor of 2, and spending per visit (if asking in monetary format).
25% of 21 to 40-year-olds are pregnant? In other words, women are expected to be pregnant for 5 out of those 20 years, which is around 6.5 Pregnancies. Sanity check your assumptions when you can!
Could you explain how the # of members subscribed is equal to the # of members visiting the gym in a given day * the frequency that they visit the gym at ? (and by the way, is that frequency monthly? I'm a bit confused). In order to find the number of members subscribed, I would multiply the number of visitors in a given day by (1/ average % of member visitors on any given day).
to validate the estimations discussed in the first part, a public health based approach would be to look up the crude birth rate (cbr) and scale it to the population of london. with a cbr of 10 across the UK for 2022, assuming that the cbr of london is identical to that of the UK, this would yield a total of 80k successful pregnancies per annum. very close!!
Thank you for useful tips and tricks of interview. I have a question about 2nd topic (daily revenues of gym). I think # of members subscribed should be calculated by this: # of members visiting the gym in a day / visit frequency (you multiplied visit frequency)
Why does it matter to break members down into peak vs non peak, members in gym at certain times, and square footage? I'm not understanding the connection in that lower level granularity. Any guidance?
Honestly,,, I don't think you need to follow it point by point. segmentation is where you bring your creativity in. if you want to segment in a simplier way, you can do so! his segments are caluclations and if you are expected to be at the top, im guessing this is the level we should be performing at? just what i think :o
It also doesn't make much sense to me to go into that much detail if we're making assumptions anyway. For example, guessing the square footage of the gym and making all those calculations vs just appointing an arbitrary number of monthly members are equally imprecise. The only reason I would think this kind of granular thinking is helpful is if the interviewer themselves gives you data about the surface area or mentions peak/off peak hours to see how you work with it.
Hello and many thanks for the video, Deniz! Very enjoyable and educational! I have a question about the first example: Are we not supposed to be dividing the number of babies born in a year by the number of babies per pregnancy and multiplying by the ratio of successful ones?
Yes, that would be ideal. I believe I've mentioned keeping it simple by assuming one baby per pregnancy and that all pregnancies are successful. If I haven't, please let me know! I definitely bring this up in our coaching sessions. In the actual interview, you can either write these points down or mention them verbally. Given the detailed nature of these components, you'll need to decide how in-depth you want to be in your approach versus managing your time effectively.
Sure, it's a bit tricky, but we're essentially aiming to understand the likelihood of a woman being pregnant. This is approached as a probability question. By dividing the total duration of pregnancies by the overall opportunity window for pregnancy (9 months * 2 / 240 months), we determine that the probability of a woman being pregnant in any given year is 7.5%.
Thanks for the video! I have a question: why wouldn't we approach the number of cars visiting gas stations daily in the last case in the following manner: 1. Multiplying the total population by the percentage of those having cars 2. (Number of days in a month) / (the average size of a car gas tank in the US / (the average number of kilometers a person makes a day / the average car gas usage in the US in liters/km)) -> how many times a month a person needs to fill his/her gas tank And then the total number of cars multiplied by that frequency will give the absolute number of servings per month / 30 = the absolute number of servings per day = number of cars visiting because they might visit the station once a day It will be less precise given the household difference and commercial cars, but most of this info should be publicly available and still reasonably averaging. I fully understand your approach which is much more precise, but if we are struggling with time on a case study, do you think my approach could work for the interviewer or it will be too large in terms of an acceptable error margin?
Thank you for your question! Generally, there are various ways to solve a market sizing question. The method you've outlined could indeed work, but it introduces an additional layer of complexity, such as average gas tank size in liters, fuel consumption in liters per kilometer, etc. As someone who doesn't have a driving license, I wouldn't be able to produce these figures on my own. Keep in mind that even if these data points are publicly available, during interviews, you're typically expected to suggest these numbers yourself. Upon reviewing your formula, I would suggest the following changes: 1) Instead of dividing "the average number of kilometers a person makes a day" by "the average car gas usage in the US in liters/km," you should multiply these values. 2) When "multiplying the total population by the percentage of those who own cars," you should also factor in the average number of cars per person to arrive at the total number of cars. My recommendation is to keep your approach simple to prevent errors! I hope this helps!
@@Prepmatter Why do you think the number of cars per person matters (vs 'car owners')? If I have 3 cars, I still can only drive 1 of them at any given time, i.e., if I make a 100km a day, and I need to visit the gas station every 700km, I will be visiting a Gas Station once a week on average irrespective of how many cars I own. Do I miss anything?
@Lagr3n, thank you for your comment! We calculate the number of cars per household, not per person, in our approach. For instance, if an average-income household owns 2 cars, it's possible that each person in the household is driving their own car. If we had defined it based on the number of cars per person, then your point would indeed be correct. I hope this clarification helps!
In estimating membership revenue in the second case, please what does the visit frequency in days help achieve? If we have 1,000 persons visiting 4 days per month, that gives a total 4,000 number of visits, not 4,000 number of gym members, no?
Okay, it seems every member wouldn't visit every single day, so we are multiplying the daily number of visits by the visit frequency in days per month to arrive at an average number of members, correct?
To understand the unique number of visitors, i.e., gym members, we first calculated the daily number of visitors. However, since these members don’t visit the gym every day, we had to consider this in our calculation. If they typically visit every 4 days (like Monday, then Friday), this means we have 4 separate groups of daily visitors. Therefore, we multiplied the number of daily visitors by their visit frequency to get the accurate count. Hope this clarifies things!
Hey there, I believe the examples under top-down and bottom-up are swapped. Top-down is Generic -> specifics and, Bottom up is Specific->Generic right!!
To make it easier to determine the appropriate approach, in top-down questions, demand is the bottleneck, whereas in bottom-up questions, supply is the bottleneck. In this video, we've covered the maternity clothing market in London, where demand is the bottleneck-meaning the larger the population of women in London, the larger the potential market. We treated the daily revenue of a local gym as a bottom-up question because we assume that a larger gym will attract more people, thus increasing revenues, assuming the demand exists. However, if the city’s population were to double, the gym’s revenues might not necessarily double due to potential constraints like space limitations. Having said that, in many market sizing questions, you can use both approaches to triangulate your results.
Hi Prep! For the third question I got 80,000 when dividing 55million by 700 off the top of my head instead of 78,600. Is this fine in a case interview or am I expected to get 78,600?
Assuming an equal age distribution and an 80-year life expectancy, each year of age represents 1/80th of the total population. Since the city's population is 8 million, we divide this by 80, resulting in 100,000 people in each year of age. Since we're looking at the number of births in a year, which corresponds to the 0-1 year age group, we can conclude that there are 100,000 births every year in the city.
Great questions! 1) Evaluating the total real estate value involves a comprehensive, top-down approach. Here are our brief insights: First, delineate the market scope you're focusing on, for example, residential, commercial, and governmental properties. Once defined, you'll need to quantify key aspects for each of these segments, including the number of units, the square meterage per unit, and the average price per square meter. For the residential segment, begin with the total number of households. For commercial real estate, estimate the number of business offices or production units, basing your calculation on the number of employed individuals. The same principle applies to governmental properties, estimating units based on employment data in such sectors. With these estimates, you can correlate the data with the number of residential flats, office spaces, and so forth. As for determining the square meterage per unit and price per square meter, consider using your flat as a benchmark, and extrapolate this data across all segments. 2) Estimating the number of leaves on a tree seems like a brainteaser; such questions are rarely asked. To tackle this question, we first need to define the scope of our inquiry - this involves determining the type of tree in question and the specific season we're observing (since leaf count can fluctuate throughout the year). The calculation essentially boils down to this formula: the number of main branches (boughs) multiplied by the number of smaller branches (twigs) per bough, multiplied by the number of leaves per twig. Given the type of tree, you should be able to assign these values reasonably. Eventhough these questions are quite rare, it's critical to approach them in a structured manner. Work alongside the interviewer to assign suitable values. Remember, these tasks are about problem-solving and demonstrating a logical approach rather than pinpoint accuracy.
![[UNCUT] ขยี้ "อลิส" ปม "คลิปลับแตงโม" เปิดชื่อคนพยายามแฮก "มือถือแตงโม" I คนดังนั่งเคลียร์I 10ก.พ.68](http://i.ytimg.com/vi/_RPLEIi5Sz8/mqdefault.jpg)
We've designed the 'Get the Offer' course to further aid your case interview preparations. This includes detailed walkthroughs of 10 more market sizing questions, live interview examples, 40+ case studies, exhibits, and structured drills. Get hands-on experience and ace your interviews! Explore more at: prepmatter.com/get-the-offer.
This is an A+ interviewee, I’m not sure how many people can just think like this on the spot
This is really good. Detailed enough to really acquire the skills. Liked, subscribed! Please keep your good work mate!
Great video, helpful content! As someone who's always been interested in management consulting, I found this really informative. Keep up the good work Deniz! I hope I'll get into the BCG Istanbul office this year :)
fingers crossed; thanks for your kind words!
Great stuff!
Excellent resource and great website. Will consider this 😮
Excellent video!
Yo, this is such an over complicated method. These are simple fermi estimation problems. No interviewer expects such minutiae, they want estimation in the correct order of magnitude. The irony is the mode you detail the the parameters, higher the risk of being wrong.
When will BCG ask amrket sizing questions? Is it part of a case or part of the fit?
Great video. I did a different approach for first case. First, I calculated total number of pregnant women at a given time point to be 375K. (6M londoners, number of 21-40yo is 1/4 so around 1.5M, out of which est 25% to be pregnant). Noting that total pregnancy duration as 40 weeks or 280 days, I decided to get rid of that factor entirely because I thought it would be plausible for anticipated mothers to buy their clothing ahead of time to make the total duration 1 year instead of 280 days for ease of math. Multipled by average number of visits to store/year as factor of 2, and spending per visit (if asking in monetary format).
Yes, this should work too (as long as you can explain some of your assumptions, such as 25% being pregnant etc.).
@@Prepmatter hey can you please make some video on guesstimate ?this is very complex
25% of 21 to 40-year-olds are pregnant? In other words, women are expected to be pregnant for 5 out of those 20 years, which is around 6.5 Pregnancies. Sanity check your assumptions when you can!
Could you explain how the # of members subscribed is equal to the # of members visiting the gym in a given day * the frequency that they visit the gym at ? (and by the way, is that frequency monthly? I'm a bit confused). In order to find the number of members subscribed, I would multiply the number of visitors in a given day by (1/ average % of member visitors on any given day).
to validate the estimations discussed in the first part, a public health based approach would be to look up the crude birth rate (cbr) and scale it to the population of london. with a cbr of 10 across the UK for 2022, assuming that the cbr of london is identical to that of the UK, this would yield a total of 80k successful pregnancies per annum. very close!!
Thank you for useful tips and tricks of interview.
I have a question about 2nd topic (daily revenues of gym).
I think # of members subscribed should be calculated by this: # of members visiting the gym in a day / visit frequency (you multiplied visit frequency)
Why does it matter to break members down into peak vs non peak, members in gym at certain times, and square footage? I'm not understanding the connection in that lower level granularity. Any guidance?
Honestly,,, I don't think you need to follow it point by point. segmentation is where you bring your creativity in. if you want to segment in a simplier way, you can do so! his segments are caluclations and if you are expected to be at the top, im guessing this is the level we should be performing at? just what i think :o
@@MiguelAnderson Thanks for the feedback!
It also doesn't make much sense to me to go into that much detail if we're making assumptions anyway. For example, guessing the square footage of the gym and making all those calculations vs just appointing an arbitrary number of monthly members are equally imprecise. The only reason I would think this kind of granular thinking is helpful is if the interviewer themselves gives you data about the surface area or mentions peak/off peak hours to see how you work with it.
Hello and many thanks for the video, Deniz! Very enjoyable and educational!
I have a question about the first example: Are we not supposed to be dividing the number of babies born in a year by the number of babies per pregnancy and multiplying by the ratio of successful ones?
Yes, that would be ideal. I believe I've mentioned keeping it simple by assuming one baby per pregnancy and that all pregnancies are successful. If I haven't, please let me know! I definitely bring this up in our coaching sessions.
In the actual interview, you can either write these points down or mention them verbally. Given the detailed nature of these components, you'll need to decide how in-depth you want to be in your approach versus managing your time effectively.
Hey. thanks for the video. Could you further explain, why we use the Total Pregnancy duration? Help is appreciated ;)
Sure, it's a bit tricky, but we're essentially aiming to understand the likelihood of a woman being pregnant. This is approached as a probability question. By dividing the total duration of pregnancies by the overall opportunity window for pregnancy (9 months * 2 / 240 months), we determine that the probability of a woman being pregnant in any given year is 7.5%.
Thanks for the video!
I have a question: why wouldn't we approach the number of cars visiting gas stations daily in the last case in the following manner:
1. Multiplying the total population by the percentage of those having cars
2. (Number of days in a month) / (the average size of a car gas tank in the US / (the average number of kilometers a person makes a day / the average car gas usage in the US in liters/km)) -> how many times a month a person needs to fill his/her gas tank
And then the total number of cars multiplied by that frequency will give the absolute number of servings per month / 30 = the absolute number of servings per day = number of cars visiting because they might visit the station once a day
It will be less precise given the household difference and commercial cars, but most of this info should be publicly available and still reasonably averaging.
I fully understand your approach which is much more precise, but if we are struggling with time on a case study, do you think my approach could work for the interviewer or it will be too large in terms of an acceptable error margin?
Thank you for your question! Generally, there are various ways to solve a market sizing question. The method you've outlined could indeed work, but it introduces an additional layer of complexity, such as average gas tank size in liters, fuel consumption in liters per kilometer, etc. As someone who doesn't have a driving license, I wouldn't be able to produce these figures on my own. Keep in mind that even if these data points are publicly available, during interviews, you're typically expected to suggest these numbers yourself.
Upon reviewing your formula, I would suggest the following changes:
1) Instead of dividing "the average number of kilometers a person makes a day" by "the average car gas usage in the US in liters/km," you should multiply these values.
2) When "multiplying the total population by the percentage of those who own cars," you should also factor in the average number of cars per person to arrive at the total number of cars.
My recommendation is to keep your approach simple to prevent errors! I hope this helps!
@@Prepmatter Why do you think the number of cars per person matters (vs 'car owners')?
If I have 3 cars, I still can only drive 1 of them at any given time, i.e., if I make a 100km a day, and I need to visit the gas station every 700km, I will be visiting a Gas Station once a week on average irrespective of how many cars I own.
Do I miss anything?
@Lagr3n, thank you for your comment! We calculate the number of cars per household, not per person, in our approach. For instance, if an average-income household owns 2 cars, it's possible that each person in the household is driving their own car. If we had defined it based on the number of cars per person, then your point would indeed be correct. I hope this clarification helps!
In estimating membership revenue in the second case, please what does the visit frequency in days help achieve?
If we have 1,000 persons visiting 4 days per month, that gives a total 4,000 number of visits, not 4,000 number of gym members, no?
Okay, it seems every member wouldn't visit every single day, so we are multiplying the daily number of visits by the visit frequency in days per month to arrive at an average number of members, correct?
To understand the unique number of visitors, i.e., gym members, we first calculated the daily number of visitors. However, since these members don’t visit the gym every day, we had to consider this in our calculation. If they typically visit every 4 days (like Monday, then Friday), this means we have 4 separate groups of daily visitors. Therefore, we multiplied the number of daily visitors by their visit frequency to get the accurate count. Hope this clarifies things!
Hey there, I believe the examples under top-down and bottom-up are swapped. Top-down is Generic -> specifics and, Bottom up is Specific->Generic right!!
To make it easier to determine the appropriate approach, in top-down questions, demand is the bottleneck, whereas in bottom-up questions, supply is the bottleneck.
In this video, we've covered the maternity clothing market in London, where demand is the bottleneck-meaning the larger the population of women in London, the larger the potential market.
We treated the daily revenue of a local gym as a bottom-up question because we assume that a larger gym will attract more people, thus increasing revenues, assuming the demand exists. However, if the city’s population were to double, the gym’s revenues might not necessarily double due to potential constraints like space limitations.
Having said that, in many market sizing questions, you can use both approaches to triangulate your results.
Hi Prep! For the third question I got 80,000 when dividing 55million by 700 off the top of my head instead of 78,600. Is this fine in a case interview or am I expected to get 78,600?
Can you explain how did you derive the 100k births a year from the 80 y life expectancy?
Assuming an equal age distribution and an 80-year life expectancy, each year of age represents 1/80th of the total population. Since the city's population is 8 million, we divide this by 80, resulting in 100,000 people in each year of age. Since we're looking at the number of births in a year, which corresponds to the 0-1 year age group, we can conclude that there are 100,000 births every year in the city.
How about total real estate value in X country or # of leaves on a tree?
Great questions!
1) Evaluating the total real estate value involves a comprehensive, top-down approach. Here are our brief insights:
First, delineate the market scope you're focusing on, for example, residential, commercial, and governmental properties. Once defined, you'll need to quantify key aspects for each of these segments, including the number of units, the square meterage per unit, and the average price per square meter.
For the residential segment, begin with the total number of households. For commercial real estate, estimate the number of business offices or production units, basing your calculation on the number of employed individuals. The same principle applies to governmental properties, estimating units based on employment data in such sectors.
With these estimates, you can correlate the data with the number of residential flats, office spaces, and so forth. As for determining the square meterage per unit and price per square meter, consider using your flat as a benchmark, and extrapolate this data across all segments.
2) Estimating the number of leaves on a tree seems like a brainteaser; such questions are rarely asked. To tackle this question, we first need to define the scope of our inquiry - this involves determining the type of tree in question and the specific season we're observing (since leaf count can fluctuate throughout the year).
The calculation essentially boils down to this formula: the number of main branches (boughs) multiplied by the number of smaller branches (twigs) per bough, multiplied by the number of leaves per twig. Given the type of tree, you should be able to assign these values reasonably.
Eventhough these questions are quite rare, it's critical to approach them in a structured manner. Work alongside the interviewer to assign suitable values. Remember, these tasks are about problem-solving and demonstrating a logical approach rather than pinpoint accuracy.
Insightful!